Embed Size (px)
Citation preview
1
The Impact of Mandatory Menu Labeling On One Fast Food Chain in King County, Washington
Eric A. Finkelstein, PhD1, Kiersten L. Strombotne, BA
1,
Nadine L. Chan, PhD, MPH2, James Krieger, MD
3
Word count (text only): 2,769
Pages: 24
Tables: 4
Corresponding Author:
Eric Finkelstein
Duke-NUS Graduate Medical School
Health Services & Systems Research
8 College Road, Level 4
Singapore, 169857
This research was funded by an internal grant from Duke-NUS Graduate Medical School. No
financial disclosures were reported by the authors of this paper.
1 Duke-NUS Graduate Medical School, Health Services & Systems Research
2 Public Health - Seattle and King County, Assessment, Policy Development, and Evaluation
3 Public Health - Seattle and King County, Prevention
2
Abstract
Background: As part of a comprehensive effort to stem the rise in obesity prevalence, King
County, Washington enforced a mandatory menu labeling regulation requiring all restaurant
chains with 15 or more locations to disclose calorie information at the point of purchase
beginning in January 2009.
Purpose: The purpose of this study is to quantify the impact of the King County regulation on
transactions and purchasing behavior at one Mexican fast food chain with locations within and
adjacent to King County.
Methods: To examine the effect of the King County regulation, a difference-in-difference
approach was used to compare total transactions and average calories per transaction between 7
King County restaurants and 7 control locations focusing on two time periods: one period
immediately following the law until the posting of drive through menu boards (January 2009 to
July 2009), and a second period following the drive-through postings (August 2009 through
January 2010). Analyses were conducted in 2010.
Results: No statistically significant impact of the regulation on purchasing behavior was found.
Trends in transactions and calories per transaction did not vary between control and intervention
locations after the law was enacted.
Conclusions: These results do not provide evidence that mandatory menu labeling positively
influenced food purchasing behavior at a single chain of restaurants.
3
I. Introduction 1
The growing prevalence of obesity over the past several decades is now well documented.1 Also 2
documented are increases in caloric intake, especially for obesity promoting energy dense foods.2 3
One reason for this increase is a trend toward consumption of food-away-from-home (FAFH). 4
Since 1972, the proportion of total food expenditures spent on FAFH increased from 34% to 5
roughly 50%.3 FAFH meals are generally higher in calories, salt, and fats than home cooked 6
meals 4 - 6
and there is evidence that increased consumption of restaurant foods, and primarily 7
fast foods, is partly responsible for rising obesity. 4, 7-10
8
9
As part of more comprehensive efforts to stem the rise in obesity prevalence, several state and 10
local governments, including New York City, San Francisco, King County Washington and 11
others have enacted or proposed mandatory menu labeling. Adoption of these regulations was in 12
part justified by evidence from experimental studies that demonstrated the provision of nutrition 13
information positively influenced choice of menu items. 11, 12
Because these laws are relatively 14
new, to date only three published studies have attempted to quantify the effect of menu labeling 15
in restaurants. The first was a 2007 study by the NYC Department of Health and Mental Hygiene 16
that examined food purchases at Subway restaurants, a chain which voluntarily posted calorie 17
information prior to enactment of the city‟s menu labeling law.13
The study found that customers 18
who looked at nutritional information prior to ordering purchased meals with fewer overall 19
calories. Although this suggests that mandatory postings may be effective, because Subway is 20
known to offer entrees seen as healthier, and many consumers may choose to eat there for this 21
reason, the extent to which results from Subway would generalize to other chains is unknown. 22
Elbel et al. studied the effect of the NYC menu-labeling law at 14 fast food restaurants in low-23
4
income, minority neighborhoods in NYC.14
They found no statistically significant effects of the 24
legislation on caloric intake. 25
26
A recent pilot study quantified the impact of voluntary menu labeling in Pierce County, 27
Washington. The study showed that the average calories purchased in 6 full-service restaurants 28
fell by about 15 calories per entrée.15
However, this study is based on only one month of data 29
post-labeling so it is unclear whether these results would be sustained. Moreover, the analyses 30
looked solely at entrees. If a customer orders a healthier entrée, it is possible that he compensates 31
with a caloric beverage or dessert. As a result, net calories could increase even if entrée calories 32
decline. In order to test the net effect of the legislation on calories purchased, all foods and 33
drinks should be included in the analysis. Based on the studies to date, it remains uncertain 34
whether or not mandatory menu labeling will lead to significant reductions in caloric intake from 35
restaurants. 36
37
This study complements prior studies by providing evidence of the impact of mandatory menu 38
labeling in King County, Washington on one fast food chain of Mexican restaurants. King 39
County includes Seattle and several outlying cities. King County‟s menu labeling law went into 40
effect on August 1, 2008, and became mandatory (fines imposed) on January 1, 2009. The 41
legislation states that restaurants that are part of chains with 15 or more outlets nationwide and 42
have annual gross sales of at least $1 million must provide nutrition labels (calories, saturated 43
fat, carbohydrates and sodium) for all standard food and beverage items at the point-of-purchase. 44
Quick-service restaurants are required to display calories on menu boards or on signs adjacent to 45
menu boards, and must make information on carbohydrate, sodium, saturated fat, and daily 46
5
recommended caloric intake readily available in pamphlets, brochures or posters. Additionally, 47
restaurants were required to post calories on drive-through menu boards beginning in August 1, 48
2009. This latter requirement is significant given that drive-through orders represent over 70% of 49
revenue for many fast food outlets.16
50
51
The King County regulation thus provides a unique opportunity to evaluate the effect of in-store 52
and drive-through menu posting on consumer behavior. Pre-post data from one regional Mexican 53
fast food chain, Taco Time Northwest, with locations within and beyond King County, were 54
used to test the impact of mandatory menu labeling on transactions and calories purchased from 55
these locations. We hypothesize that as a result of the legislation: 56
Total transactions at locations within King County will decrease after the legislation 57
goes into effect compared with locations outside King County. This hypothesis is 58
based on the assumption that some consumers of high calorie entrees, upon disclosure 59
of the calorie information, will opt to dine at other establishments. 60
Average calories per transaction will also decrease relative to non-King County 61
locations as some consumers switch to lower calorie food and drink options in efforts 62
to reduce their caloric intake. 63
The effects of the policy will be greater after August 1, 2009 when calorie 64
information appears on drive-through menu boards. 65
These results may provide useful information for the development of the federal menu-labeling 66
law, details of which are still being considered. 67
68
II. Data and Methods 69
6
This analysis is based on sales data from Taco Time Northwest restaurants. Of the chain 70
restaurants we contacted, Taco Time was the only quick-service chain that agreed to provide 71
transaction data. Taco Time Northwest is a Mexican fast food restaurant chain with over 70 72
locations across the state of Washington. The menu includes a variety of Tex-Mex options like 73
burritos, tacos, salads and fries. Menu items span a wide range of calories. For example a beef 74
Roma burrito is 843 calories while a regular chicken taco salad is 196 calories. Notably, Taco 75
Time highlights several low-calorie entrée options on their “healthy highlights” menu. 17
76
77
The store-level data includes total monthly transactions and monthly sales for every menu item 78
between January 2008 and January 2010, 13 months after the law was enacted and the menu 79
boards were updated to include the nutritional information. Fourteen stores were included in the 80
analysis. These include all 7 stores located in counties adjacent to King County and whose data 81
were available in the company database for the entire analysis period and a randomly selected 82
subset of 21 King County stores that also had complete sales and transactions data over the study 83
period. Monthly sales for each menu item were converted to monthly calories sold based on 84
calorie data for each menu item available from the company‟s website and, for a few 85
discontinued or non-standard items, directly from company management. The resulting dataset 86
provides over 80% power to detect differences of 25 calories or more per monthly transaction as 87
a result of the legislation. 88
89
To examine the effect of the menu-labeling law on calories per transaction, a difference-in-90
difference regression was estimated focusing on two time periods: one period immediately 91
following the law until the posting of drive through menu boards (Post-period 1: January 2009 to 92
7
July 2009), and a second period following the drive-through postings (Post-period 2: August 93
2009 through January 2010). The baseline period included monthly values for January 2008 94
through December 2008. The difference-in-difference regression allows for comparing the 95
changes in 1) transactions and 2) calories per transaction between the pre- and each post period 96
within King County, and to assess whether or not these changes are larger than changes in 97
locations outside King County, where menu labeling was never implemented. Specifically, the 98
regressions were estimated in the following form: 99
100
yst = β0 + β1KC + β2 POST1 + β3(KC*POST1) + β4POST2 + β5(KC*POST2)t + εst 101
102
where yst is the dependent variable for each store s in month t (either transactions or calories per 103
transaction), KC is a dummy variable equal to one if store s is located in King County, POST1 is 104
a dummy variable for period 1 (equal to 1 when month t falls between January 2009 through July 105
2009), POST2 is a dummy variable for period 2 (equal to 1 when month t falls between August 106
2009 through January 2010). The interaction terms, KC*POST1 and KC*POST2 test the key 107
hypotheses, that pre-post changes in average monthly transactions and average monthly calories 108
per transaction are different in KC locations than in surrounding locations as a result of the menu 109
labeling legislation and drive through postings. Negative coefficients on these variables are 110
consistent with the primary hypotheses of smaller growth (or larger reductions) in these 111
outcomes. Seasonal dummy variables were included to control for temporal effects. 112
Regressions were estimated for total calories per transactions, and separately for food and drink 113
calories. Regressions that include monthly dummy variables and that break the POST1 and 114
8
POST2 time periods into smaller increments to test for temporary effects of the calorie postings 115
were also explored. Results (available upon request) were robust to all specifications modeled. 116
117
Additional analyses were run to test whether those who frequented KC locations may have been 118
making healthier purchases before the law took effect. Lower calorie food options were 119
identified as the „healthy highlights‟ listed on the company‟s menu and website. On average, 120
these entrees were 42% lower in calories than other entrees. “Healthy Highlights” entrees 121
showed a mean number of calories of 281, compared to a mean of 480 in all taco, burrito and 122
salad entrees. The lower calorie drink options included diet sodas, water, iced tea and other low-123
calorie drink options. 124
125
All analyses were conducted using Stata, version 11. Standard errors for all regression analyses 126
were adjusted for repeated observations within restaurants over time. 127
128
III. Results 129
Table 1 compares the results from the pre- and post-periods and shows no statistically significant 130
trend in monthly transactions for either King County or non-King County locations. The table 131
reveals that the number of monthly transactions per store is, on average, greater in KC than in 132
non-KC locations both before and after the legislation went into effect. Table 1 also presents 133
results for 1) average monthly calories per transaction, 2) average monthly food calories per 134
transaction, and 3) average monthly drink calories per transaction. Average calories per 135
transaction are roughly 180 calories greater in the non-King County, compared with the King 136
County locations, both before and after the menu labeling law went into effect. This difference is 137
9
largely driven by lower average food calories per transaction (roughly 160 calories lower) but 138
also by lower drink calories per transaction (roughly 20 calories higher) in King County 139
locations. King County locations show slight increases in overall calories and in calories from 140
food and slight decreases in drink calories between the pre- and each post period, with no change 141
in total calories. Non-King County restaurants show slight decreases in average drink calories 142
per transaction only. Although these differences are statistically significant, they are extremely 143
small. 144
145
Tests of the key hypotheses are summarized by the difference-in-difference estimates in Table 2. 146
These estimates are not statistically different from zero, suggesting the hypothesis that the 147
legislation did not reduce calories per transaction (either before or after calorie information was 148
posted on the drive-through menu boards) could not be rejected. 149
150
Table 3 compares the sales mix across King County and non-King County locations. These 151
results reveal no statistically significant differences in the mix of sales across major categories of 152
foods/drinks. However, table 4 shows that King County consumers were making healthier 153
purchases prior to enactment of the law. The percentage of transactions that involved „healthy 154
entrees‟ were 11.7% in King County versus 9.4% in the non-King County locations. Moreover, 155
whereas 45.4% of transactions involved a low calorie drink in King County, this figure was 156
39.4% for restaurants outside King County. These differences, which were statistically 157
significant, explain why average calories per transaction were greater in stores outside King 158
County and may explain the lack of effect of the legislation; King County patrons were already 159
consuming healthier options. 160
10
161
IV. Discussion 162
The results for this chain of Mexican fast food outlets show no statistically significant impact of 163
mandatory menu labeling on monthly transactions and calories sold per transaction as 164
implemented in King County, Washington. Neither total monthly transactions nor calories per 165
transaction were immediately impacted by the legislation or impacted later when calorie 166
information was added to the drive-through menu boards. 167
168
Given the pending federal legislation, it is important to consider possible explanations of the lack 169
of effectiveness of the King County legislation at this chain. One possible explanation is that 170
customers were already aware of the calorie content of the menu items. This is possible for Taco 171
Time and for most other fast food outlets as this information is almost universally available on 172
the company websites. If consumers are already aware of the calorie content of fast food menu 173
items, then posting this information on the menu boards is likely to have little added value. 174
175
Although this explanation is plausible, numerous studies have shown that consumers tend to be 176
poor judges of the caloric content of restaurant foods, and infrequently access web-based 177
nutrition information.18, 19
Therefore, having the information on the menu boards is likely to 178
convey new information to consumers. However, it is possible that consumers do not understand 179
or internalize the information on the menu board and/or the link between a poor diet, obesity, and 180
adverse health outcomes. If consumers are unable to understand or internalize the menu postings, 181
then mandatory menu labeling without an accompanying public health or education campaign is 182
unlikely to be successful. This should be an area for future research. 183
11
184
It is also possible that, even when confronted with all relevant health and nutrition information at 185
the point of purchase, taste, price convenience and variety remain more salient factors in the 186
purchasing decision for many consumers than do the potentially adverse health effects of 187
consuming a particular menu option. Understanding the extent to which select subgroups of 188
consumers are willing to trade off taste, price, convenience, and variety for improved health 189
content should also be an area for future research. 190
191
It is worth noting that even before the King County law went into effect, on average, customers 192
of the King County locations were eating healthier than customers outside King County. Seattle 193
is known to be a health conscious city so this result is not surprising. However, it raises the 194
question of how customers identified the healthier entrees. It is possible that the more health 195
conscious consumers went to the website for this information, but more likely is that they relied 196
on information available at the point of purchase. This information included the „healthy 197
highlights‟ logo displayed on the menu board and also the identification of drinks as ‟diet‟ or 198
„sugar-free‟. All are strong cues for which are the lower calorie menu options and suggest that 199
these types of logos may be as effective or perhaps even more effective than detailed nutrition 200
information in encouraging healthier food purchases. This too should be an area for future 201
research. 202
203
This analysis has several limitations. The primary limitation is that it is limited to one fast food 204
chain with 13 months of data post legislation. And although the results are generally consistent 205
with the lack of calorie reduction seen from previous studies, future studies should replicate 206
12
these results for other establishments and over longer time periods before strong conclusions can 207
be made concerning the overall impact of mandatory menu labeling. Ideally, these studies will 208
include several of the largest national fast food chains. An additional limitation of this analysis is 209
that because it was at the store level, it was not possible to identify whether certain subgroups 210
(e.g., more health conscious, parents ordering for children, or those with chronic illnesses who 211
are more motivated to choose healthier options) differentially benefited from the legislation. 212
Future studies allowing for subgroup analyses would be beneficial. 213
214
As noted above, there were differences in average calories per transaction between King County 215
and non-King County stores prior to enactment of the legislation. As a result, it is possible that 216
the non-King County stores were not appropriate controls. However, the lack of any statistically 217
significant reduction in transactions, and statistically significant increases, as opposed to 218
decreases, in calories per transaction in King County stores, suggests that contamination is 219
unlikely to be masking a real positive effect of the legislation. Finally, this analysis focused 220
solely on demand responses to menu-labeling legislation. Future studies should examine the 221
extent to which mandatory menu-labeling encourages supply-side changes and their subsequent 222
impact on fast food purchases. Supply side effects may involve changes in-store promotions, 223
product mix or reformulation of existing products. 224
225
Conclusions 226
These results do not provide evidence that mandatory menu labeling, as implemented in King 227
County, Washington positively influenced food purchasing behavior at one type of fast food 228
chain. In lieu of the pending federal legislation, future qualitative and quantitative studies should 229
13
be undertaken to identify the circumstances under which mandatory menu labeling is likely to be 230
most effective. 231
232
14
Acknowledgements 233
The authors greatly appreciate the data and assistance provided by the management at Taco Time 234
Northwest. 235
236
15
References 237
238
1. Flegal KM, Carroll MD, Ogden CL, Curtin LR. Prevalence and trends in obesity among US 239
adults, 1999-2008. JAMA. 2010;303(3):235-41. 240
2. Finkelstein EA, Strombotne K. The economics of obesity. Am J Clin Nutr. 2010;91(5):1520S-4. 241
3. Rydell SA, Harnack LJ, Oakes JM, Story M, Jeffery RW, French SA. Why eat at fast-food 242
restaurants: reported reasons among frequent consumers. J Am Diet Assoc. 2008;108(12):2066-243
70. 244
4. Lin B, Frazao E, Guthrie J. Away-from-home foods increasingly important to quality of 245
American diet. Washington DC: U.S. Department of Agriculture, 1999 (Agric Info Bull 749):1-246
22 247
5. Young LR, Nestle M. The contribution of expanding portion sizes to the US obesity epidemic. 248
Am J Public Health. 2002 February 1, 2002;92(2):246-9. 249
6. Young LR, Nestle M. Expanding portion sizes in the US marketplace: implications for nutrition 250
counseling. J Am Diet Assoc. 2003;103(2):231–234 251
7. French S, Harnack L, Jeffery R. Fast food restaurant use among women in the Pound of 252
Prevention study: dietary, behavioral and demographic correlates. Int J Obes Relat Metab 253
Disord. 2000;24(10):1353 - 9. 254
8. Niemeier HM, Raynor HA, Lloyd-Richardson EE, Rogers ML, Wing RR. Fast food 255
consumption and breakfast skipping: predictors of weight gain from adolescence to adulthood in 256
a nationally representative sample. J Adolesc Health. 2006;39(6):842-9. 257
9. Pereira MA, Kartashov AI, Ebbeling CB, Van Horn L, Slattery ML, Jacobs JDR, et al. Fast-food 258
habits, weight gain, and insulin resistance (the CARDIA study): 15-year prospective analysis. 259
Lancet. 2005;365(9453):36-42. 260
10. Thompson O, Ballew C, Resnicow K, Must A, Bandini L, Cyr H, et al. Food purchased away 261
from home as a predictor of change in BMI z-score among girls. Int J Obes Relat Metab Disord. 262
2004;28(2):282 - 9. 263
11. Ippolito, PM, Pappalardo, JK. Advertising, nutrition, and health: evidence from food advertising 264
1977-1997. Bureau of Economics Staff Report, Federal Trade Commission. 2002. 265
12. Variyam, JN, Cawley, J. Nutrition labels and obesity. NBER working paper no. 11956. 2006. 266
16
13. Bassett M, Dumanovsky T, Huang C, Silver L, Young C, Nonas C, et al. Purchasing behavior 267
and calorie information at fast-food chains in New York City, 2007. Am J Public Health. 268
2008;98(8):1457-9. 269
14. Elbel B, Kersh R, Brescoll V, Dixon L. Calorie labeling and food choices: a first look at the 270
effects on low-income people in New York City. Health Aff. 2009;28(6):w1110-21. 271
15. Pulos E, Leng K. Evaluation of a voluntary menu-labeling program in full-service restaurants. 272
Am J Public Health. 2010:100(6):1035-9. 273
16. Hughlett M. Drive-thoughts done right ring up returns. Chicago Tribune [Internet]. 2008 Nov 28 274
[cited 2010 July 13]; Available from: http://articles.chicagotribune.com/2008-11-275
28/news/0811270365_1_drive-through-restaurant-technologies-competitive-advantage 276
17. Taco Time full nutritional guide. 2010 [updated 2010 July; cited 2010 Aug 24]. Available from: 277
http://www.tacotimenw.com/tacotimemenu.aspx 278
18. Burton S, Creyer EH, Kees J, Huggins K. Attacking the obesity epidemic: the potential health 279
benefits of providing nutrition information in restaurants. Am J Public Health. 2006;96(9):1669-280
75. 281
19. Krukowski RA, Harvey-Berino J, Kolodinsky J, Marsana RT, Desisto TP. Consumers may not 282
use or understand calorie labeling in restaurants. J Am Diet Assoc. 2006;106(6):917-20. 283
284
17
List of Figure Titles 285
Table 1. Unadjusted mean differences in transaction data (per store, per month) 286
Table 2. Difference-in-difference regression results and standard errors 287
Table 3. Differences in overall purchasing behaviors in the pre-period 288
Table 4. Differences in healthy purchasing behaviors prior to the menu-labeling law 289
18
Table 1. Unadjusted mean differences in transaction data (per store, per month)
Pre-period Post-period 1 Post-period 2
Difference
between post-
period 1 and
pre-period
Difference
between post-
period 2 and
pre-period
King County
Average monthly transactions 11,592.3 11,766.5 11,001.3 174.2 -590.9
Average calories per transaction 1,211.3 1,217.0 1,214.3 5.7* 2.9*
Avg food calories per trans 1,127.6 1,136.0 1,134.8 8.4* 7.2*
Avg drink calories per trans 83.8 81.0 79.5 -2.7* -4.3*
Non-King County
Average monthly transactions 10,193.6 10,258.4 9,823.2 64.7 -370.5
Average calories per transaction 1,391.4 1,392.3 1,375.8 0.9 -15.6
Avg food calories per trans 1,289.0 1,293.5 1,279.3 4.5 -9.7
Avg drink calories per trans 102.4 98.8 96.5 -3.6* -5.9*
* Implies difference is statistically significant (p< 0.05).
19
Table 2. Difference-in-difference regression results and standard errors
Difference-in-
difference Post-
period 1
Difference-in-
difference Post-
period 2
Average calories per transaction 4.8 18.5
(7.86) (15.11)
Avg food calories per trans 3.9 16.9
(7.11) (13.55)
Avg drink calories per trans 0.9 1.7
(1.19) (1.64) a There were no statistically significant differences in results.
b Seasonal dummy variables (winter, spring, summer) were included as covariates.
20
Table 3. Differences in overall purchasing behaviors in the pre-period
Pre-Period
KC Non-KC Difference
Entrees sold (% of all items sold) 47.0% 47.3% -0.4%
Drinks sold (% of all items sold) 28.0% 27.2% 1.0%
Desserts sold (% of all items sold) 1.0% 1.0% -0.1%
Sides sold (% of all items sold) 21.4% 21.6% -0.3%
Kidmeals sold (% of all items sold) 2.6% 2.8% -0.2%
100.0% 100.0% a There were no statistically significant differences in results.
21
Table 4. Differences in healthy purchasing behaviors prior to the menu-labeling law
Pre-Period
KC Non-KC Difference
Healthy Entrees (% of all entrees) 11.7% 9.4% 2.3%*
Diet Drinks (% of all drinks)
45.4%
39.4%
6.0%*
* Implies difference is statistically significant (p< 0.05).
22
Appendix for Reviewers
. reg trans kc period2 period3 kcpost1 kcpost2 winter spring summer, cluster(id)
Linear regression Number of obs = 350
F( 8, 13) = 148.54
Prob > F = 0.0000
R-squared = 0.1354
Root MSE = 2342.3
(Std. Err. adjusted for 14 clusters in id)
------------------------------------------------------------------------------
| Robust
trans | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
kc | 1398.619 1240.14 1.13 0.280 -1280.541 4077.779
period2 | -177.8657 210.1731 -0.85 0.413 -631.917 276.1857
period3 | 176.1496 186.4996 0.94 0.362 -226.7584 579.0575
kcpost1 | 109.4702 281.7799 0.39 0.704 -499.2781 718.2186
kcpost2 | -220.4476 236.8633 -0.93 0.369 -732.1596 291.2644
winter | -409.5024 79.58915 -5.15 0.000 -581.4443 -237.5605
spring | 861.802 102.5654 8.40 0.000 640.2228 1083.381
summer | 1078.981 130.3144 8.28 0.000 797.4534 1360.508
_cons | 9810.823 625.715 15.68 0.000 8459.048 11162.6
------------------------------------------------------------------------------
. reg calpertrans kc period2 period3 kcpost1 kcpost2 winter spring summer, cluster(id)
Linear regression Number of obs = 350
F( 8, 13) = 39.07
Prob > F = 0.0000
R-squared = 0.5161
Root MSE = 86.683
(Std. Err. adjusted for 14 clusters in id)
------------------------------------------------------------------------------
| Robust
calpertrans | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
kc | -180.0403 45.12396 -3.99 0.002 -277.5247 -82.55595
period2 | -7.298437 5.825349 -1.25 0.232 -19.88334 5.286464
period3 | -3.372899 14.72485 -0.23 0.822 -35.18401 28.43821
kcpost1 | 4.784758 7.868881 0.61 0.554 -12.21493 21.78444
kcpost2 | 18.53234 15.11442 1.23 0.242 -14.12038 51.18506
winter | 27.7701 5.183897 5.36 0.000 16.57097 38.96923
spring | 26.80574 6.421267 4.17 0.001 12.93344 40.67804
summer | 38.801 6.293257 6.17 0.000 25.20524 52.39676
_cons | 1368.027 38.28348 35.73 0.000 1285.321 1450.734
------------------------------------------------------------------------------
23
. reg foodcalpt kc period2 period3 kcpost1 kcpost2 winter spring summer, cluster(id)
Linear regression Number of obs = 350
F( 8, 13) = 48.04
Prob > F = 0.0000
R-squared = 0.4838
Root MSE = 83.095
(Std. Err. adjusted for 14 clusters in id)
------------------------------------------------------------------------------
| Robust
foodcalpt | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
kc | -161.4088 43.29315 -3.73 0.003 -254.938 -67.87968
period2 | -3.744679 5.069139 -0.74 0.473 -14.69589 7.206529
period3 | 2.5953 13.0374 0.20 0.845 -25.57028 30.76088
kcpost1 | 3.916135 7.11211 0.55 0.591 -11.44864 19.28091
kcpost2 | 16.88139 13.55264 1.25 0.235 -12.3973 46.16009
winter | 27.96627 4.527917 6.18 0.000 18.1843 37.74824
spring | 28.3554 5.679606 4.99 0.000 16.08535 40.62544
summer | 37.52268 5.660483 6.63 0.000 25.29395 49.75141
_cons | 1265.513 36.467 34.70 0.000 1186.731 1344.295
------------------------------------------------------------------------------
. reg drinkcalpt kc period2 period3 kcpost1 kcpost2 winter spring summer, cluster(id)
Linear regression Number of obs = 350
F( 8, 13) = 34.72
Prob > F = 0.0000
R-squared = 0.7383
Root MSE = 5.611
(Std. Err. adjusted for 14 clusters in id)
------------------------------------------------------------------------------
| Robust
drinkcalpt | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
kc | -18.63151 2.600335 -7.17 0.000 -24.24919 -13.01383
period2 | -3.553755 1.094095 -3.25 0.006 -5.917404 -1.190105
period3 | -5.968203 1.729639 -3.45 0.004 -9.704862 -2.231544
kcpost1 | .8686238 1.186628 0.73 0.477 -1.694931 3.432179
kcpost2 | 1.650949 1.644919 1.00 0.334 -1.902684 5.204581
winter | -.1961831 .8002945 -0.25 0.810 -1.925114 1.532748
spring | -1.549652 .8913469 -1.74 0.106 -3.47529 .375986
summer | 1.278317 .7413702 1.72 0.108 -.3233157 2.87995
_cons | 102.5146 2.393879 42.82 0.000 97.3429 107.6862
------------------------------------------------------------------------------
