# Questions tagged [covariate-balance]

The tag has no usage guidance.

13 questions
Filter by
Sorted by
Tagged with
12 views

### Summary and plot after MatchIt for specific patient subgroup

I have used Matchit::matchit() to implement 1-to-1 propensity score matching on a dataset and cobalt::love.plot to show covariate balance for the dataset before and after adjustment. Is it possible to ...
28 views

### Use SMD or raw difference in proportions when comparing balance of binary covariates used in propensity matching?

I am a junior in the medical field, and have been tasked with performing a propensity matched analysis of some data, using R. I am looking at comparing the balance of my covariates (both before and ...
25 views

### Issues calculating SMDs after applying IPTW weights

I get the exact same standardized mean differences (SMDs) before and after applying inverse probability of treatment (IPTW) and stabilized IPTW weights to my data. I reproduced this below using the ...
179 views

### How can restricted randomization to achieve covariate balance lead to imbalance in unobserved variables?

In literature, designing an experiment is considered a trade-off between covariate balance and robustness. For example Harshaw et al. (2024) writes In an effort to make the estimators more precise, ...
• 683
1 vote
24 views

### After matching: How do I interpret the value of the type ‘distance’ (=Propensity score) in the balance measures table of the r-package cobalt bal.tab?

I have used the R-package ‘MatchIt’ to perform (1) a nearest neighbour propensity score matching (NNM) based on the Framingham Heart Study and (2) for comparison, an optimal PS matching (OM) for the ...
29 views

### What to do with large standardized mean difference in ipw analysis

I'm working on a cohort data, and I'm trying to evaluate the influence from loss to follow-up and missing values. So I performed multiple imputation with chained equation first, then use the imputed ...
• 137
1 vote
15 views

### Checking covariate balance in survey response

In survey adjustment for nonresponse, a model-based approach can model response propensity on a variable where each observation is a respondent or nonrespondent from the full sample. An approach to ...
• 111
96 views

### Is it ok to include unbalanced covariates in regression after full matching?

A couple of my covariates remained imbalanced after full matching. Is it acceptable to include those imbalanced covariates in the post-matching regression analysis (in addition to the matching weights)...
30 views

### Using exact same covariates in CBPS model and regression model

When using CBPS in R and following Imai and Ratkovic, do you necessarily need to include all the covariates in your covariate-balancing procedure that you are planning to use in the subsequent ...
• 101
1 vote
378 views

### Difference between normalized difference and standardized mean difference in cobalt?

In Imbens & Wooldridge (2009, p. 19), they define the normalized difference as: whereas the cobalt's package standardized mean difference uses by default (for the ATE) "the 'pooled' standard ...
• 187
326 views

### Why is covariate balance fundamental for causal inference in observational data but unnecessary in experimental data?

The observational case For observational data, Hernán & Robins (2023, p. 49) state: In the absence of marginal randomization, [computing the average causal effect in the entire population] ...
• 1,484
123 views

### Evaluating success of propensity score matching with single metric that accounts for both covariate balance and matching rate?

Propensity score matching techniques can be assessed and compared with covariate balance metrics like the standardized mean difference (SMD). However, SMDs don't account for varying matching rates. ...
• 6,164
1 vote
It is known that a two-factorial design is balanced if and only if each combination of the level of the two factors have the same replication. As an example, if $A$ has $2$ levels and $B$ has $2$ ...