week 13: hypothesis testing type 1 Flashcards
how to find a viable research topic
- what are your professional interests
- read, read, read
- use clinical experinces
- identify significant questions that were posed and investigated but not answered in a definite manner (there was a small sample size, uncontrolled variables, etc)
ideas to build on existing research
- expand research to a new age group/diagnostic group
- apply different outcome measures
- address questions of social validity
- change the stimuli used
- change the setting of evaluation/treatment
- use up-to date instrumentation
variables
phenomena you plan to observe
independent variable
conditions or manipulations being studied
*what you change
dependent variable
observations or measures obtained
*what you measure
what will a well-written research question include
- includes all variables in an unambiguous way
- it should also include the intent of the research is it to
- –describe persons or circumstances
- -discover relationships between variables
- –identify differences between groups
ways to formulate a research problem
- does not have to be in the form of a question or series of questions
- can be
- –formal hypothesis
- –statement of purpose
- –conditional if-then statement
hypothesis
a formal statement of predicted outcome of the study
Ho
null hypothesis
*based on the assumption that the results will yield no significant differences and/or relationships between variables
Ha
alternative hypothesis
- statement of what the researchers expected to find when they conducted the study
- –directional is stated in a way in which the researchers had reason to believe a particular outcome would happen (such as a positive relationship between 2 variables)
- –non-direction is states in a way that a significant relationship is expected, but no indication of positive or negative effect
two decision options for a hypothesis test
reject the null hypo meaning it is not valid or
fail to reject the bull hypo meaning it is valid
statement of purpose
highly flexible
*explains the focus of nearly any type of research including descriptive, relational, and difference studies
conditional if-then statements
if the findings turn out one way, the study supports a certain conclusion, but if they findings turn out another way, the study supports a different conclusion
what are the criteria for well-informed research questions
- operationalize: give precise, specific details in the description of variables
- experimental treatments needs to be one that other professionals can carry out in a similar way
- –variables also need to be defined in a way that leads to valid and reliable conclusions
what are the 6 steps of hypothesis testing
1) state the hypothesis
2) set a level of risk
3) choose a sample size
4) determine the critical value
5) compute the test statistic
6) accept or reject the null hypothesis
type 1 error
reject Ho when it is true
type 2 error
fail to reject Ho when it is false
which type of hypothesis error is considered more serious
type 1 error
what is the typical risk level set at for hypothesis testing
0.05 or less which means we are willing to risk a type 1 error 5/100 times
what does the sample size determine for hypothesis testing
the probability distribution to be used
- z statistic requires large samples over 30
- t statistic is appropriate for smaller samples under 30 if certain assumptions are met
the power of a test
the probability of rejecting the null hypothesis when it is false
- the larger the sample, the more powerful the test
- conclusions drawn from small sample sizes have a high probability of accepting the null hypothesis when it is false (type 2 error)
the critical value
the cutoff, it is found be using tables of normal distributions, student’s t, etc
*it depends on the alternative hypothesis and the alpha level
what are the two types of alternative hypothesis
- two-tailed meaning nondirectional
- –predicts a treatment effect but not specifying which is more effective
- onetailed or directional
- –hypo the treatment will be better than the other
- ***should generally be two tailed unless there is amazing evidence in favor or a specific direction
computing the test statistic
a standard score such as a z or t statistic
what factors does the choice of test statistic depend on
- the hypothesis
- the distribution of the target population
- the sample size
- other special characteristics of the sample
non parametric statistics
distribution free *a special class of inferential stats that depends on the ranking
making a decision about the null hypothesis
- test statistic is compared to the critical value
- reject the Ho supports the Ha but does not prove it
- we never accept the Ha
nominal level of measurement
*researchers assign participants and their responses to categories
*cannot be ranked or arranged in order
(numbers can be arbitrarily assigned like 1 for female and 2 for male)
*cannot compare the categories directly (one is not more or less than the other)
*effective nominal measures are exhaustive meaning every participant fits a category
, and mutually exclusive so only fit one category
ordinal level of measurement
- rank ordered data (suggests inequality of data and lacks other numerical properties
- can be from high-low relative one another or use a rating scale to assign number (likert scale)
- gives info about the positions but not the difference between them
- can use labels instead of numbers
interval level of measurement
which participants have higher or lower scores as well as how much they differ
- data has arithmetic properties of inequality and equal intervals
- examples: IQ and hearing threshold
- cannot compare scores directly because does not have a true zero
ratio level of measurement
*all of the features of an interval, with a true zero (all arithmetic properties)
*3 characteristics
1)ability to arrange numbers on a continuum
20 ability to specify amount and difference between amounts
3) ability to identify an absolute zero relative to a characteristic
*often measures a physical attribute: actual height, intensity duration
tables
visual representation of numerical data about participants or statistical findings of a study
pie chart
appropriate for nominal, categorical measures
*useful for illustrating percentages or proportions of observations that fit into particular categories
histograms vs bar graphs
useful for illustrating the magnitude or frequency of one or more variables
- histograms are for continuous data
- bar graphs are used with categorical data
- –often used for depicting group differences on measures like frequency counts, means, etc
line graph
lke bar/column graphs, it is useful for illustrating the magnitude or frequency of one or more variables
- –particularly suitable for depicting several values in a series or for depicting a special kind of nonlinear relationship called an interaction
- –interaction occurs when two or more groups respond in unique ways to the experimental manipulations
box and whiskers plot
displays the medial and upper quartile (75th %) and lower quartile (25%)
*easy way to show outliers
scatter plot
appropriate for illustrating the relationship between two (sometimes three) continuous measures
—bivariate