Experimental Design and the struggle to control threats to validity

Research questions and hypotheses Research questions and hypotheses  A research question is a problem or a statement that is in need of a solution or answer.  A hypothesis is a proposed answer to the research question.

Research Hypotheses  Unambiguously identify and describe a relationship between two or more variables.  Are empirically testable.

For example  H1: Fee increases greater than $5 per visit will result in a 10 percent reduction in use by visitors with family incomes below $25,000.  Null Hypothesis (H 0 ): There is no difference in impact of fees....

Break the question down into its sub-components Do families with young children make more use of city parks than families with college- aged children?  Lacks Specificity  Needs Simplification  Provides no Direction Who uses parks?  Vague  A research IDEA

The sub-components of a question  Do families with young children  make more use of  city parks  than families with college-aged children?  Who, #1 = SUBJECT  What =CONCEPT  Where= CONTEXT  Who, #2 = SUBJECT This gives us an idea of what we can observe or measure (what DATA we will be collecting), from whom, and where.

Research questions and hypotheses are design- specific  What do people in Moscow do when it snows a lot here?  Why do some people in Moscow not know how to drive in snow?  Is the amount of skill a person has driving in snow related to the length of time that a person has lived in a place with snow?

and..  CSS students are more skilled snow drivers than ECB students.  A Ford Taurus with an automatic transmission will handle worse in snow than will an identical car with a manual transmission (regardless of driver characteristics, skill level and experience).

Experimental Design

NATURALISTIC CASE-STUDY CORRELATIONAL DIFFERENTIAL EXPERIMENTAL INCREASINGLY CONSTRAINED LOW HIGH

Experimental research allows us to test hypotheses and infer causality under controlled conditions designed to maximize internal validity. The high control and internal validity often mean a reduction of external validity. That is, the more precise, constrained, and artificial we become in the experiment, the less natural are the procedures and findings. The result is that we sometimes have difficulty generalizing experimentation to the natural environment.

Experimental research allows us to test hypotheses and infer causality under controlled conditions designed to maximize internal validity. The high control and internal validity often mean a reduction of external validity. That is, the more precise, constrained, and artificial we become in the experiment, the less natural are the procedures and findings. The result is that we sometimes have difficulty generalizing experimentation to the natural environment.

Experimental design is a planned interference in the natural order of events by the researcher.

Experimental  Comparisons are made under different and controlled conditions.  Subjects are assigned to each type of condition in an unbiased manner, usually matched or random.  Although causality can often be inferred, results may not be applicable outside of the experimental setting.

CONTROL TREATMENT #1 TREATMENT #2TREATMENT #3 TIME 0: PRE-TEST. Collect baseline data. Experimental

CONTROL JOLT COKECOFFEE TIME 1: TREATMENT GIVEN Experimental

CONTROL JOLT COKECOFFEE TIME 3: POST-TEST. Collect data on the effects of the treatment and compare to pretest, and to each treatment. Experimental

Hypotheses  NULL (Statistical) mean JOLT= mean COKE = mean COFFEE = mean CONTROL  RESEARCH (Causal) Caffeine causes people to grow tall, except when combined with sugar.  CONFOUNDING (Rival) Genetic differences caused the observed differences in response.

A simple 2-group, posttest-only design  Outfitters given low-impact training  Outfitters given NO low-impact training  Measure impacts caused by their clients Compare Scores

Hypotheses  NULL (Statistical) mean Trained= mean Un-trained  RESEARCH (Causal) The clients of outfitters trained in low impact methods will cause fewer impacts than clients of outfitters who did not receive such training.  CONFOUNDING (Rival) Prior knowledge may have caused the observed differences in response.

Threats to Validity

Validity  Because of the confounding Hypothesis we are not 100% sure that our conclusions are valid.  Did we indeed measure the effects of new knowledge?  Independent -- Dependent?  Other Variable/s -- Dependent?

Road map for research success  Anticipate all threats to validity.  Take plans to eliminate them.  Statistical validity  Construct validity  External validity  Internal validity.

Statistical validity  The variations in the dependent variable are not due to variation in the subjects, but due to variations in the measuring instrument. The instrument is unreliable.  Some statistical assumptions have been violated (e.g., non-normal data treated with parametric statistics; means of ordinal data, etc.).

Construct validity  How well the observed data support the theory, and not a rival theory.

External validity  The degree to which the findings of the study are valid for subjects outside the present study. The degree to which they are generalizable.  Unbiased, complex sampling procedures; many studies, mid-constraint approaches help strengthen external validity.  internal

Internal Validity  Threats to causality (our ability to infer).  Did the independent variable cause the dependent to change (were they related), or did some confounding variable intervene?  Maturation, history, testing, instrumentation, regression to the mean, selection, attrition, diffusion and sequencing effects.

Maturation  If the time between pre- and posttest is great enough to allow the subjects to mature, they will!  Subjects may change over the course of the study or between repeated measures of the dependent variable due to the passage of time per se. Some of these changes are permanent (e.g., biological growth), while others are temporary (e.g., fatigue).

History  Outside events may influence subjects in the course of the experiment or between repeated measures of the dependent variable.  Eg., a dependent variable is measured twice for a group of subjects, once at Time A and again at Time B, and that the independent variable (treatment) is introduced between the two measurements.  Suppose also that Event A occurs between Time A and Time B. If scores on the dependent measure differ at these two times, the discrepancy may be due to the independent variable or to Event A.

Testing  Subjects gain proficiency through repeated exposure to one kind of testing. Scores will naturally increase with repeated testing.  If you take the same test (identical or not) 2 times in a row, over a short period of time, you increase your chances of improving your score.

Instrumentation  Changes in the dependent variable are not due to changes in the independent variable, but to changes in the instrument (human or otherwise).  Measurement instruments and protocols must remain constant and be calibrated.  Human observers become better, mechanical instruments become worse!

Regression to the mean  If you select people based on extreme scores (High or low), in subsequent testing they will have scores closer to the mean (they would have regressed to the center).

Selection  When random assignment or selection in not possible the two groups are not equivalent in terms of the independent variable/s.  For example, males=treatment; females=control.  Highest threats in naturalistic, case study and differential approaches.

Attrition  When subjects are lost from the study.  If random it may be OK.  Confounding attrition is when the loss is in one group or because of the effects of the independent variable. (Jolt killed off 2 people!)

Diffusion of treatment  When subjects communicate with each other (within and between groups) about the treatment) they diffuse the effects of the independent variable.

Sequencing effects  The effects caused by the order in which you apply the treatment.  A B C  A C B  B A C, etc.

Subject effects  Subjects “know” what is expected of them, and behave accordingly (second guessing).  Social desirability effect.  Placebo effect. A placebo is a dummy independent effect. Some people react to it.

Experimenter effects  Forcing the study to produce the desired outcome.  Expectations of an outcome by persons running an experiment may significantly influence that outcome.

Single- and double-blind procedures  Single blind –subjects don’t know which is treatment, which is not.  Double blind—experimenter is also blind.

Designs  One shot: –G1T------O2  One Group Pre-Post: –G1O T------O2  Static Group: –G1 T------O2 –G2 (C) O2

More designs  Random Group: –RG1T O –RG2 (C) O  Pretest-Posttest, Randomized Group: –RG1O T------O2 –RG2 (C) O O2

Yet another design:  Solomon four-group: –RG1O T------O2 –RG2O O2 –RG3 T O2 –RG4 O2

One last one !  Latin Square: –RG1OT1OT2OT3O –RG2OT2OT3OT1O –RG3OT3OT1OT2O