Ch4 objectives

Designing a Study Objectives and Hypotheses The following chapter is excerpted from Designing HIV/AIDS Intervention Studies: An Operations Research Handbook, Andrew Fisher and James Foreit, 2002, Washington, DC: Population Council. (More on OR Handbook)

../orhndbk.html

C H A P T E R

OBJECTIVES AND

HYPOTHESESThe objectives and hypotheses of a research study should flow logicallyfrom the earlier sections identifying the problem situation, defining theparameters of the problem, and justifying its importance. In thissection, we explain how to narrow and focus the research. Specificobjectives are written that describe the expected results arising from thestudy and the outcome variables that will be measured. Once objec-tives have been set, researchers can formulate specific, testable hypoth-eses that specify the relationship between program interventions andoutcomes.

Ultimate Objectives

Most research studies include a statement of ultimate objectives thatdescribes the expected implications or contributions arising from thestudy. The ultimate objective of many HIV/AIDS studies is to preventthe transmission of HIV and/or mitigate the impact of AIDS. This isan ambitious goal. It is unlikely that any single study will have asignificant effect on reducing HIV transmission or a major effect onmitigating the impact of AIDS. A somewhat more reasonable ultimateobjective is that the study will provide program administrators andpolicymakers with information useful for improving HIV/AIDSprograms, scaling up these programs, or developing more effectivepolicies.

Ultimate objectives relate the reasonable and expected contributions ofthe study to broad social, economic, or health concerns. In this way,the ultimate objectives contribute to the justification of why researchon the problem is required. Note, however, that ultimate objectives areexpected contributions. In a sense, they represent wishful thinking.The investigator does not promise that the contributions will occurand therefore usually does not try to measure them.

4

2626 DESIGNING HIV/AIDS INTERVENTION STUDIES

Immediate Objectives

In contrast to ultimate objectives that state what isexpected to happen, immediate objectives statewhat will happen. Immediate objectives relatedirectly to the research problem situation. Theyindicate the variables that will be examined andmeasured. An immediate objective represents apromise by the investigator that certain activitieswill take place and specific variables will be exam-ined.

Whenever possible, the immediate objective shouldbe stated in behavioral terms—in other words, interms of what actions or behaviors will take place.The immediate objective should specify who willdo, how much of what, to whom, when, where,and for what purpose.

Examples of Immediate Objectives• “Over a 24-month period, the National HIV/

AIDS Prevention Council in Uganda, togetherwith Makerere University, will develop ayouth-friendly program to provide comprehen-sive voluntary counseling and testing six daysof every week to youth ages 14 through 24.This program will be evaluated in terms of thenumber of youth who use the service, the cost-effectiveness of the service, and the reductionof unsafe sexual risk behaviors by youth.”

• “Over a three-year period, the provincial healthdepartment, together with two local NGOs,will test and compare in terms of reduced STIsand in terms of cost-effectiveness two ap-proaches to implementing a 100 percentcondom use policy in brothels. One approachwill be tested in Puerto Plata and the other inSanto Domingo. At each site, all commercialsex establishments will be enrolled in the study.All sex workers and brothel owners will receiveextensive group counseling on the need to usecondoms. Educational materials will be madewidely available and condoms will be distrib-uted free of charge.”

Examples of Ultimate Objectives• “The ultimate objective of this intervention

study is to contribute toward improving thequality of life for PLHA and the quality andeffectiveness of prevention, care, and supportactivities offered by community-based organi-zations to PLHA.”

• “The ultimate objective of this interventionstudy is to develop a cost-effective model ofservice delivery in Kenya that can reduce thetransmission of HIV from mothers to theirnewborn children.”

• “The ultimate objective of this study is todevelop for Uganda a cost-effective and sus-tainable approach for integrating high-qualityVCT services with related family planning andSTD and TB prevention and care services inprimary health care facilities.“

• “The ultimate objective of this interventionstudy is to reduce the transmission of HIV andother STIs by implementing a 100 percentcondom use policy in brothels throughout theDominican Republic.”

• “The ultimate objective of this interventionstudy is to promote in Cambodia the conceptof dual protection against STIs and pregnancythrough the use of male and female condoms.”

Comments on the ExamplesEach of these ultimate objectives briefly states theoverall expected outcome of the study. Theseobjectives tend to be broad in nature. They repre-sent the expected impact the researchers hope thestudy will have. Ultimate objectives relate thepurpose of the study to larger health concerns, suchas improving the quality of life for PLHA, reduc-ing the rate of mother-to-child transmission(MTCT), or integrating HIV/AIDS services withother health care services.

2727OBJECTIVES AND HYPOTHESES

tance by youth of VCT, cost-effectiveness, sexualrisk reduction, reduction of STIs, reduced trans-mission of HIV from mother to child, and in-creased condom use.

Each of these variables needs to be carefully definedin operational terms later in the proposal, but at leastas stated here in the immediate objectives, theyindicate the focus of the proposed research and someof the key variables that will be examined.

Hypotheses

A hypothesis is a statement about an expectedrelationship between two or more variables thatpermits empirical testing. While ultimate objec-tives identify the anticipated contributions arisingfrom a study, and immediate objectives (stated inbehavioral terms) specify what will be done ormeasured in the study, hypotheses specify theexpected relationship among the variables. Hy-pothesis statements are most appropriate for fieldintervention or evaluative studies. Diagnostic orexploratory studies do not normally requirehypothesis statements because they generally donot test relationships between variables.

Study hypotheses serve to direct and guide theresearch. They indicate the major independent anddependent variables of interest. They suggest thetype of data that must be collected and the type ofanalysis that must be conducted in order to mea-sure the relationship among the variables.

A single hypothesis might state that variable A isassociated with variable B, or that variable Acauses variable B. Sometimes a hypothesis willspecify that, under conditions X, Y, and Z,variable A is associated with or causes variable B. Awell-written hypothesis focuses the attention of theresearcher on specific variables.

• “Over a three-year period, researchers atKenyatta National Hospital will develop andimplement a comprehensive training programfor staff, and an HIV/AIDS voluntary counsel-ing and testing service for pregnant womenattending the antenatal unit at the hospital.Women who test positive will be offeredantiretroviral medication plus breast milksubstitutes. The effect of these activities will bemeasured in terms of reduced rates of mother-to-child HIV transmission, cost-effectiveness,acceptability to staff, and participation bypregnant women and their partners.”

• “Over a four-year period, the Amber andVictoria Gold Mining Company, togetherwith three local NGOs, will implement acomprehensive STI control program amongminers and sex workers. This program willconsist of behavior change communicationusing peer educators, condom promotion anddistribution through social marketing, and STIservices, including periodic presumptivetreatment among sex workers. The effect ofthese activities will be measured in terms ofreduced prevalence of STIs among miners andsex workers, greater condom use, increased safesex practices, and increased knowledge aboutHIV/AIDS transmission.”

Comments on the ExamplesNote that, in each of these objectives, the investiga-tor promises to undertake a specific activity: todevelop a program activity and measure the effect;to test and compare two different program strate-gies for implementing a 100 percent condomprogram; to develop, implement, and measure theeffect of an MTCT program; and to implementand measure the effect of an STI reduction pro-gram among miners and sex workers. The situationor context involved in the research is also specified.Who will conduct the study, where it will beconducted, and when it will be conducted arestated. Finally, and most important, the keyvariables of interest are indicated, such as accep-


When writing hypothesis statements, it is impor-tant to keep in mind the distinction betweenindependent and dependent variables. An inde-pendent variable causes, determines, or influencesthe dependent variable. An example of the basicrelationship between these two types of variables isshown in Figure 4.1. This model shows a directrelationship. In other words, whenever the inde-pendent variable changes, the dependent variablechanges. The dependent variable depends on theindependent variable.

This type of direct relationship is usually the modelused to develop study hypotheses, but in somecases, a study will hypothesize a model in whichthe relationship is indirect. The independentvariable acts on the dependent variable throughintervening variables. These intervening variablesserve to either increase or decrease the effect theindependent variable has on the dependent variable.

Intervening variables are sometimes referred to astest or control variables. In longitudinal fieldexperiments, the design of the study (with randomassignment of cases to experimental and compari-son areas) usually controls the intervening variables.In cross-sectional surveys, intervening variables aremeasured and then controlled in the statisticalanalysis. Many HIV/AIDS behavioral and epide-miological research studies are based on a modelthat includes intervening variables (see Figure 4.2).

FIGURE 4.2Interveningvariables

FIGURE 4.1A direct relationshipbetween indepen-dent and dependentvariables

Independentvariables

Dependentvariables

Cause,determine,or influence

Independentvariables

Interveningvariables

Dependentvariables

Cause,determine,

or influence

For example, in an effort to increase condom use(the dependent variable), an HIV/AIDS programmight initiate a large information, education, andcommunication campaign (the independentvariable). Alone and by itself, an IEC programcannot increase condom use. The program mustact on and operate through a set of interveningvariables, which in turn cause, determine, orinfluence condom use. There may be many ofthese intervening variables, but the most likely onesthat might be influenced by an IEC programwould be people’s knowledge about HIV transmis-sion, their attitudes about the use of condoms,their sexual risk behaviors, and their beliefs abouttheir vulnerability to AIDS. A possible researchmodel for an evaluation study of the effects of anIEC program is shown in Figure 4.3.

Acting

on an

d

opera

ting

thro

ugh


FIGURE 4.3The effects of interveningvariables in an IEC program

In HIV/AIDS operations research, the hypothesesof most importance usually are those that considerprogram activities as the independent variable.The HIV/AIDS program input, such as an IECeffort, a voluntary counseling and testing service,an STI prevention program, or a condom promo-tion program, is the independent variable. Theobjective of the research is to determine the extentto which the independent variable affects thedependent variable. The dependent variable isusually the central concern of a research proposal’sproblem statement. The intervening variables areimportant because they tend to either increase ordecrease the strength of the relationship betweenthe program (the independent variable) and theoutcome effect (the dependent variable). It istherefore important to gather information onpossible intervening variables.

HIV/AIDS IECprogram

Condom use

Cause,

determine,

or influence

Most research studies will examine many indepen-dent variables and many intervening variables butonly a few dependent variables. In writing studyhypotheses, always think in terms of the expectedrelationship between variables. Think first aboutthe central problem your study will address (thedependent variable). Next, consider what factor orfactors (the independent variables) might cause,determine, or influence the dependent variable.Finally, ask yourself if the relationship between theindependent and dependent variables is direct orindirect through a set of intervening variables.

Examples of Hypotheses• Students who participate in a school-based life

skills education program will have moreknowledge about HIV risk behaviors andprevention practices and more positive atti-tudes about safe sex practices than comparablestudents who do not participate in the lifeskills program. As a result, they will be morelikely to delay sexual debut and use condomsthan comparable students who do not partici-pate in the life skills program.

• PLHA who receive comprehensive counselingon antiretroviral therapy that includes discus-sion of the side effects and their managementbefore ARV treatment begins are more likely toadhere effectively to treatment after a year thanPLHA who have not received counseling onARVs before they begin therapy.

• Peer educators who receive a five-week, field-based training course will have higher knowl-edge about HIV/AIDS and be more highlymotivated than peer educators who havereceived a three-week, classroom-based trainingcourse. As a result, work performance of thepeer educators who have been trained for fiveweeks will be significantly higher than peereducators who have received a three-weektraining course that was based in the classroom.

Acting

on an

d

opera

ting

thro

ugh

People’s knowledge,attitudes, beliefs,

behaviors


• Higher levels of peer educator work perfor-mance will lead to higher rates of consistentcondom use among youth.

• Voluntary counseling and testing programs thatare linked to community-based HIV/AIDSorganizations will be more successful in reduc-ing stigma and discrimination directed atPLHA than VCT programs not linked tocommunity-based HIV/AIDS organizations.

• Reduced levels of stigma and discriminationagainst PLHA will lead to more people seekingvoluntary counseling and testing.

• Community-based HIV/AIDS organizationsthat actively involve PLHA in the planning andimplementation of programs will be morelikely to achieve their objectives than similarorganizations that do not involve PLHA in theplanning and implementation of programs.

• Dual protection programs that focus oncounseling women together with their malepartners will be more successful than dualprotection programs that focus only on coun-seling women.

Comments on the ExamplesNote that in each hypothesis there is a statement ofan expected relationship between two or morevariables. In the first hypothesis, the expectedrelationship is between attending a life skillseducation program (the independent variable) andknowledge about HIV/AIDS, and more positiveattitudes toward safe sex (the intervening variables).These intervening variables are then hypothesizedto have a relationship with the delay of sexualdebut and on condom use (the dependent vari-ables). In the second hypothesis, the relationship isbetween comprehensive counseling on ARVs (theindependent variable) and adherence to ARVtreatment after a year (the dependent variable).

The third and fourth hypotheses go together.In the third hypothesis, the relationship is betweena five-week, field-based training program (theindependent variable) and greater knowledge aboutHIV/AIDS and higher motivation (the interveningvariables). These intervening variables are thenhypothesized to have an effect on work perfor-mance (the dependent variable). Hypothesis 4carries this causal process one step further bysuggesting that work performance (which is nowthe independent variable) will have an effect oncondom use (the dependent variable). Takentogether, the model for these two hypotheseswould look like Figure 4.4.

FIGURE 4.4A model for two hypotheses about the effectof a peer educator training programon higher rates of condom use

(Hypothesis 3)

(Hypothesis 4)

Higher workperformance

Causes,determines,or influences

Highercondom use

Acting

on an

d

opera

ting

thro

ugh

Cause,determine,

or influence

Higher knowledgeand motivation

Peer educatortraining program

Higher workperformance


Similarly, hypotheses 5 and 6 go together. Inhypothesis 5, VCT programs linked to commu-nity-based organizations (the independent variable)are hypothesized to reduce stigma and discrimina-tion against PLHA (the dependent variable). Inhypothesis 6, reduced stigma and discriminationbecome the independent variable, and use of VCTcenters becomes the dependent variable.

For all the hypotheses listed in the examples, theimportant point is that they show a relationshipbetween an independent and a dependent variable,and it is this relationship that is tested empiricallythrough a research study.

To summarize, all research studies should include astatement of (1) ultimate objectives, (2) immedi-ate objectives, and (3) hypotheses (except in thecase of purely exploratory or descriptive studies).Each of these statements serves a different purpose:

1. Ultimate objectives state the anticipated(hoped-for) contributions of the study.

2. Immediate objectives state what will be donein the study.

3. Hypotheses state the expected relationshipbetween two or more variables.


1. List the major variables of your study underthe headings: Independent Variables andDependent Variables. Intervening variables canbe listed under both headings. The dependentvariable(s) should relate directly to yourproblem statement.

2. Review the list of dependent variables and thenwrite a statement of ultimate objectives thatrelates to them. Ask yourself, “If I knew thefactors that caused, determined, or influencedthe dependent variable(s), how would thisknowledge help policymakers, programadministrators, or others?” How will the resultsfrom the study help improve service delivery,improve training programs, or assist in thedesign of educational materials? In otherwords, what are the anticipated contributionsof the study?

3. Now write the immediate objectives for yourstudy. Focus your attention on the specificactions that you will perform. Exactly what doyou plan to do? What variables do you plan tomeasure? Write behavioral objectives thatanswer the following questions:

• Who will do?• How much of what?• To whom?• When?• Where?• For what purpose?

4. Finally, write the hypotheses that your studywill test. Review once again your list ofindependent and dependent variables. How arethese variables related? Are there special condi-tions that must be present before they arerelated? Write hypothesis statements in posi-tive, not negative, terms. Write a hypothesisfor each major relationship that you expect totest in your study.

What To Do: Writing Ultimate Objectives, Immediate Objectives, and Hypotheses

Health & Medicine

Ch4 objectives