3583072
Description
Flashcards by seonapalmer, updated more than 1 year ago
|
Created by seonapalmer
over 8 years ago
|
|
| Question | Answer |
| What is Structural Equation Modelling? | Structural equation modeling (SEM) is a statistic method in analysing the associations between variables and the sequential ordering of these variables. |
| Why would you use structural equation modelling? | * To answer complex questions * Researcher can make predictions about multiple dependant variables. * Researchers can hypothesise how items load onto factors *Researcher can model latent (unobserved) variables. |
| Can SEM test for causation? | No, unless your data is longitudinal |
| What are the statistical analyses that fall under the umbrella of SEM? | *Path Analysis * Confirmatory Factor Analysis (CFA) * Structural Equation Model (SEM) * A mix of two (PA & CFA) |
| Can SEM model error (Unexplained Variance)? | Yes, which means we can estimate how much variance is actually being explained. We can identify unexplained and explained variance for everything in our model. |
| What is the key question that you ask yourself in SEM? | How well does your model (Your hypotheses) fit the data collected? |
| What is path analysis? | Path analysis is a straightforward extension of multiple regression. Its aim is to provide estimates of the magnitude and significance of hypothesised causal connections between sets of variables. This is best explained by considering a path diagram. |
| How do we evaluate how well our model fits the data? | Goodness of fit test (Chi Square) Observed -Expected. The goal is that the values between the two models will be as close as possible. Therefore Chi square should be non significant. |
| Do we want a difference between the observed and expected correlation matrixes? | No, the less difference between the two, the better the model fit and therefore the more variance that it accounts for. |
| Is the Chi square analysis conducted on the variance - covariance matrix or the correlation matrix | Variance -covariance matrix (Unstandardised). |
| The Maximum likelihood Chi square is the most common chi square in SEM, path analyis and CFA, what does it mean? | It’s trying to mazimise the likelihood that what you have hypothesised will map onto whats going on in your data. (to what extent does the data fit my implied/hypothesised model?) |
| Why are there instances where the model fits the data well, yet the chi square is significant? and what do you do about it? | * The chi square test can be quite sensitive partiularily when sample sizes increase, trivial differences show as statistically sig. (the problem is that SEM works best with large samples). This can also happen as the model becomes for complex. * To counter this use fit indices |
| What are fit indices? | They are used to assess fit of model. |
| What are the two types of fit indices? | * Absolute fit indices * Incremental fit indices (Generally researchers will use both forms of indices to evaluate model fit in addition to Maximum Likelihood Chi Square. |
| What is the absolute fit indices? | How well the hypothesied model fits the sample data, *SRMR *RMSEA |
| What is incremental fit indices? | The fit of the hypothesised model compared to the fit of the null model (a model where the variables are specified not to correlate). So we are looking at to what extent our model does a better job of explaining the data than a model that has nothing correlated. * CFI *TLI |
| What are the guidelines for interpreting the CFI and TLI? | Values range from 0 - 1. Above .90 represents acceptable fit however above .95 is regarded as good fitting. |
| What are the guidelines for interpreting RMSEA and SRMR? | We want these to be as low as possible. RMSEA: Values less that .08 or less suggest reasonable approximation of error. SRMR: Values .08 or less represent a well fitting model with little error. |
| What is an Endongenous variable? | Dependent variable generated within a model and, therefore, a variable whose value is changed (determined) by one of the functional relationships in that model. |
| If a model is a poor fit we can examine a series of statistics to understand the source(s) of misfit. What is this called | Specification Search - this can inform researcher of the need for post hoc modifications, adding, deleting paths etc. |
| Post hoc respecifications can only be made if there is some type of? | Theoretical reasoning |
| Misfit can be identified by looking at these sources? | * Modification indices * Expected parameter change (EPC) * The standard residual covariance matrix |
| What are the two ways we can make changes to the path analysis model? | Model Building Model Trimming |
| What is the significance test conducted to determine the effect of adding or deleting a pathway? | Chi Square difference test *generally in adding or deleting a pathway you are either adding or reducing or increasing the DF by 1. |
| What are the assumptions of path analysis? | * Independence of observations * Normal Distribution – populations from which samples are taken are normally distributed (bell shaped curve). |
Want to create your own Flashcards for free with GoConqr? Learn more.