Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.

Normal Size Small Size show me how

Normal Size Small Size show me how

# Methods Part II

Question | Answer |
---|---|

Experimenter Bias | Researchers beliefs about the expected outcome influence the results (leading questions, inadvertant reinforcement) |

Threats of Experimenter Bias | Internal Validity, it affects results. External Validity, you can't generalize to natural settings. |

Demand Characteristics | Aspects of the study or study environment that reveal the hypothesis being tested...may lead subjects to exhibit subject role (good, negativistic, apprehensive, faithful) |

Threats of Participant Bias | Internal, innaccurate results. External, can't generalize. ( use deception, single blind study, control group) |

Sampling Distribution of the means | Permute data to gather all possible samples of n size. Take mean of each possible permuted sample and build distribution. |

Single Sample T-test | Evaluates sample mean against sampling distribution mean. (population data is known) |

Basic Units of a Sampling Distribution | Xbar. Muxbar. SigmaSQRDxbar. Sigmaxbar <- std error. |

Standard Error | Standard Deviation of a sampling distribution |

Central Limit Theorem | Specifies nature of sampling distribution. *mean of sampling distribution is a pretty good estimate of the pop. mean for samples larger than N=1. Sampling distributions are more normal, with less variability. |

Why T-test? | Z underestimates population variance and gives too many rejections of null. T has more variability, flatter (platykurtic) |

Degrees of Freedom | # of Observations that are free to vary (last has to make dataset have a the set Xbar) |

Two Sample T-test (Independent) | Compares two means from two groups (usually 1 IV w/ 2 levels). |

Independent T-test Notation | Xbar1, Xbar2, S(xbar1-xbar2) |

Sampling Distribution for Independent T-test | Sampling distribution of differences between the 2 sample means. |

Variance Sum Law | The variance of the sampling distribution is the sum of the variances for the component sampling variances (i.e. std dev = S(xbar1-xbar2)) |

Pooled Standard Deviation | Assume equal variances in an Independent T-test, we factor variance out (still under radical) |

Assumptions for an Independent (Two Sample) Ttest | Independent Random Sampling. Normal Populations. Equality of Variance. DV is ratio or interval. |

Intact Groups | Groups pre-formed because variable being study cannot be randomly assigned. forces btwn subjects design. May affect validity of test |

Confidence Intervals | Obtained sample mean(s) +- (TCRIT*STD ERROR). std error bars will be smaller than Confidence interval |

Paired Sample T-test (Dependent) | ALL ABOUT DIFFERENCES. within subjects. Difference score from each individual tested against expected difference of 0. |

Between Subjects ANOVA | a multi-group generalization of the t-test w/ 3 distinctions: more groups, focus on variance instead of means, uses F. Same assumptions as t |

ANOVA is one tailed because | F=t^2. Also made of SS/df and SUM of squared values cannot be negative and df cannot be negative. |

Hypothesis for Btwn Sbjt ANOVA | All are equal. Two are different from one another. NON DIRECTIONAL |

Indications of SSbtwn & SSwithin | If SSbtwn is large & SSwithin is small, null is probably true. If btwn is a fair amount and within is somewhat less, alternative is probably true. (because F=between/within) |

Family-Wise Error Rate | Refers to the chance of committing at least one type-1 error among a set of analyses |

Fisher's LSD Test | Least Significant Difference. We run modified tests between pairs ONLY IF ANOVA is significant. pretty liberal. |

Bonferroni | Alpha adjustment technique. Alpha family wise is divided by total # of comparisons. |

Post Hocs | Use DFwithin to get Tcrit from post hoc comparisons (ttests) |

Problems with Btwn Subjects ANOVA | Individual Differences cause high within group variability and mask treatment effect. Individual differences can also become confounding variables |

Subjects variability | Makes up part of Within group variability in One-way anova. Tells how much within groups variability can be attributed to individual differences |

One Way ANOVA | more sensitive to treatment effect because individual differences are accounted for. We use means of groups and means of individual subjects |

Factorial Design | A research study involving 2 or more IVs. |

Advantages of Factorial Designs | More realistic...DVs of interest are rarely ever effected by only 1 thing in real life. Shows interactions of IVS on DVs. Economical, can test multiple hypotheses at once. |

Main Effect | The mean differences among the levels of 1 factor |

Theres an Interaction if | The effects of one factor depend upon the level of another factor. If it is significant we can no longer talk about main effects. |

Simple Main Effect | The effect of one factor at one particular level of another factor |

Null Hypotheses for Factorial ANOVA | All treatments for factor A are equal. All treatments for factor B are equal. Factors are independent. |

Factorial ANOVA | we split SSbtwn(called SScells) into 3 groups. A, B & Interaction btwn A&B. Indent three groups if including Fcells in source table |

Testing Single value against known sample or population | Z test |

Mean of one group against population mean | Single sample t-test |

More than one independent variable | Use factorial ANOVA test |

One IV, two levels, Between Subjects | Independent Ttest |

One IV, two levels within subjects | Dependent T-test |

One IV, three+ levels, within subjects | One way, repeated measures ANOVA |

One IV, three+ levels, between subjects | One way, between subjects anova |

One Way ANOVA means | One IV, multiple levels |

Created by:
lpicklesimer