Lecture Conducting and reading research in health and human performance (4/e): Chapter 4 - Ted A. Baumgartner, Larry D. Hensley

Chapter 4 - Developing the research plan. The following will be discussed in this chapter: Research approaches, hypotheses, types of hypotheses, hypothesis testing, direction of expected results, data collection techniques,...

Chapter 4 Developing the Research Plan

Important to most research studies

Tentative explanation of the outcome of a

– based on theory or previous research

Null or Statistical Hypothesis - A hypothesis of

“no difference or no relationship”

– primary use is for statistical testing

– hypothesis which says the independent variable has no effect on

Hypothesis Testing

The Research Hypothesis is transformed into

a Statistical or Null Hypothesis (Ho)

– This is done so that statistical tests can be

employed that will determine whether the findings are statistically significant or can be attributed to chance

– The results of the statistical test will enable the

researcher to accept or reject the null hypothesis

More Hypothesis Testing

The purpose of the statistical test is to evaluate the null hypothesis at a specified level of

probability

– For instance, testing the difference in the mean values between

2 groups at the 05 level means:

Do the values of the dependent variable differ significantly (p<.05) so that these differences would not be attributable to chance occurrence more than 5 times in 100?

– If the null hypothesis is accepted, then the researcher rejects the research hypothesis and concludes there is no difference

between the groups

Example Research Hypothesis

It is hypothesized that children taught by

teaching method A will perform better on a

reading achievement test than children taught by method B

Direction of Expected Results

Directional Hypothesis – when the researcher has reason to believe a particular relationship or

or difference exists in any direction

– There is a difference in the motivational level of

Example Null Hypothesis

There will be no significant difference in reading performance between students taught by

method A and students taught by method B

orTeaching method has no effect on the reading performance of students

Rule of Thumb

Research which asks DIFFERENCE or

RELATIONSHIP questions should always have hypotheses

Research which asks DESCRIPTIVE questions (with no comparisons across groups) may not need hypotheses

You can only reject or fail to reject the null

hypothesis

Data Collection Techniques

The nature of the study will determine what type

of data are required to answer the question and the method of collecting these data

Multiple techniques may be used in a single

study

Three Basic Techniques

Observation - the researcher may watch the

research participants perform and record

relevant information about them

Measurement - the researcher may test the

research participants or apply a device to

measure certain qualities

Questioning - the researcher may ask the

research participants questions to obtain

information

Observation Techniques

Direct observation

Indirect observation

Participant observation

Indirect Observation

Research participants are filmed or videotapedResearcher views tape

Participant Observation

The observer participates in the research

setting with the research participants, often

spending considerable time in the natural setting developing field notes

Qualitative research methodology

Measurement Techniques

This broad category of techniques involves

actively testing the research participants on the characteristics of interest

Almost anything can be measured

Major categories of measures include

– Physical

– Cognitive

– Affective

Scaling Techniques

Scaling is the process of assigning numbers to the various levels of a particular concept that we wish to measure Thus, a scale provides an

indirect measure of the concept of interest

Scales can be used to obtain information on

almost any topic, object, or subject Attitude,

opinion, behavior, performance, and perception are frequently measured by some type of scale

Common Scales

Rating Scale

Semantic Differential Scale

Rank Order Scale

Likert Scale

Rating Scale

Individual items are judged on a single

dimension and scored on a linear scale or

continuum by selecting a numerical or verbal point on the scale that corresponds to their impression of the item

Numerical Rating Scale

Numerical Rating Scale

Verbal Rating Scale

Importance Importance Importance

Semantic Differential Scale

Semantic Differential Scale

A scaling method designed for measuring ones

“image” of a selected topic or concept Subjects will choose a relative position between pairs of bipolar adjectives which describe the topic along

a single dimension

No more than about 20 items should be used Responses are converted to numeric values and treated statistically

Semantic Differential Scale

Semantic Differential Scale

Rank Order Scale

Items are ranked, usually in terms of preference

or importance, relative to each other This

forced ranking results in ordinal scores, thus

limiting the statistical treatment of the scores

The number of items to be ranked should be less than 10, to avoid making the task too difficult

Rank Order Scale

Please rank the brands of beer listed below in order of  preference, with a 1 being the brand you most prefer, 2  being your second choice, and so forth.

  Budweiser   Coors

  Miller   Corona

Likert Scale

A very popular scaling technique which

measures the respondent’s degree of agreement

or disagreement on an issue, opinion, or

particular belief

The continuum of response typically runs from

SA , A, U, D, to SD

Responses to a Likert scale can be considered

to be interval level scores, thus allowing scores

to be summed and treated statistically

Likert Scale

Pick a number from the scale to show how much  you agree or disagree with each statement:

Scales & Statistics

Some controversy exists among researchers

and statisticians regarding the appropriate

statistical treatment of scaled responses

– If the intervals between score points are presumed to

be equal, thus resulting in interval data, the responses can be analyzed statistically

Interval Level Scales

Most authorities are willing to accept Likert

Scales, Semantic Differential Scales, and to a lesser extent Rating Scales, as meeting these assumptions, thus enabling the scores to be treated statistically

Ordinal Level Scales

On the other hand, there is virtually no

controversy concerning Rank Order Scales such

as “Forced Ranking Scales” or “Paired

Comparison Scales”

Responses constitute ordinal data thus it is inappropriate to perform arithmetic operations or combine responses for a total scale score

Report frequencies and percentages only

May be mailed, distributed by the researcher or completed online

Many formats

Interviews

Essentially an oral questionnaire

May be personal or telephone interviews

– Structured interview

– Unstructured interview

Focus Group Interview

Essentially an interview with groups of peopleDesigned to stimulate participants free

expression of feelings, beliefs, etc

Requires a skilled facilitator to guide discussion

Delphi Technique

Unique questioning method used to get

consensus on a specific issue or topic

Involves obtaining responses from a well-defined group of individuals

Each person then reviews his/her position based upon the collective responses from the group

and revises position as warranted

May require several iterations

Selecting the Data Collection Method

What type of data is needed to answer the

research problem?

Factors to consider

– Demands on the research participant

– Costs in terms of money, energy, and time

– Ability of the researcher to handle the selected

technique, including the data analysis

Data Collection Instruments

May include any mechanical or electronic

equipment, physical performance task, and-pencil test or scale, as well as a

paper-questionnaire designed to collect data on the variable of interest

Researcher’s choice of instrument involves

deciding if one already exists that can be used

as is, if one exists but needs to be revised, or if one needs to be developed

Instrument Selection

Thoroughly review the literature

If instrument is found, assess suitability

– Reliability - consistency with which it measures

– Validity - measures what it is suppose to measure

Reliability and validity of an instrument are often specific to the age, gender, characteristics of subjects

on which it is used Without acceptable reliability and validity, the data are

of no use in answering the research question

Instrument Revision

If an instrument is found, but it is not quite

acceptable for the current research situation, it may be modified or revised

Permission should be obtained before revising copyrighted instrument developed by someone else

If changes are major, then it may be necessary

to determine new indices of reliability and validity

of the revised instrument

Instrument Development

This is a time-consuming and difficult task

Only undertake if there is no existing instrument that will suffice

Basic steps include the following:

1 Review the literature

2 Develop tentative instrument

3 Obtain opinions of experts concerning the instrument

4 Revise the instrument as needed

5 Pilot test the instrument

6 Further revise the instrument as needed